Effect of rapamycin on lysosomal accumulation in a CRISPR/Cas9‐based cellular model of VPS13A deficiency

Abstract VPS13A is a lipid transfer protein localized at different membrane contact sites between organelles, and mutations in the corresponding gene produce a rare neurodegenerative disease called chorea‐acanthocytosis (ChAc). Previous studies showed that VPS13A depletion in HeLa cells results in an accumulation of endosomal and lysosomal markers, suggesting a defect in lysosomal degradation capacity leading to partial autophagic dysfunction. Our goal was to determine whether compounds that modulate the endo‐lysosomal pathway could be beneficial in the treatment of ChAc. To test this hypothesis, we first generated a KO model using CRISPR/Cas9 to study the consequences of the absence of VPS13A in HeLa cells. We found that inactivation of VPS13A impairs cell growth, which precludes the use of isolated clones due to the undesirable selection of edited clones with residual protein expression. Therefore, we optimized the use of pool cells obtained shortly after transfection with CRISPR/Cas9 components. These cells are a mixture of wild‐type and edited cells that allow a comparative analysis of phenotypes and avoids the selection of clones with residual level of VPS13A expression after long‐term growth. Consistent with previous observations by siRNA inactivation, VPS13A inactivation by CRISPR/Cas9 resulted in accumulation of the endo‐lysosomal markers RAB7A and LAMP1. Notably, we observed that rapamycin partially suppressed the difference in lysosome accumulation between VPS13A KO and WT cells, suggesting that modulation of the autophagic and lysosomal pathway could be a therapeutic target in the treatment of ChAc.

in this organism. 10 Consistent with this hypothesis, delayed digestion of phagosomal contents is observed in Tetrahymena thermophila mutant cells lacking a VPS13A homologue, leading to impaired phagocytosis. 11 Furthermore, accumulation of p62 and protein aggregates is observed in the central nervous system of Drosophila lacking the VPS13A functional homologue. 12 This hypothesis has been reinforced by studies of human red blood cells and reticulocytes from ChAc patients, which showed organelle remains and delayed clearance of mitochondria and lysosomes, respectively. 13 Furthermore, Vps13a −/− mice showed impaired autophagy in the basal ganglia with accumulation of toxic levels of the active kinase Lyn. 14 All these results suggest that the autophagic/lysosomal pathway could constitute a potential therapeutic target for the treatment of ChAc.
The availability of simple cellular models is essential to identify compounds of therapeutic interest. Previous studies showed that VPS13A depletion in HeLa cells results in deficient lysosomaldependent degradation of autophagic and endocytic cargos, such as p62, LC3 and EGFR. 5,10 A partial defect in the maturation of the lysosomal hydrolase cathepsinB was also described. 5 This impaired degradative capacity of lysosomes affects autophagy and endocytic degradation at a late stage, resulting in the perinuclear accumulation of vesicles containing partially degraded material (endolysosomes) as determined previously by electronic microscopy. 5 This phenotype can be easily monitored by immunofluorescence analysis of endosomal and lysosomal markers. 5,10 Our goal was to optimize a HeLa cell model of VPS13A dysfunction based on these previous observations to determine, as a proof of concept, whether compounds that modulate the endo-lysosomal pathway could be beneficial in the treatment of ChAc. To test this hypothesis, we use pooled cells obtained shortly after transfection with CRISPR/Cas9 components that generate a mixture of wild-type and VPS13Aedited cells, allowing a comparative analysis of phenotypes and avoiding the selection of clones with residual levels of VPS13A.
Consistent with previous results obtained by siRNA treatment, VPS13A inactivation by CRISPR/Cas9 results in accumulation of the endo-lysosomal marker LAMP1. 5 We found that rapamycin, a potent activator of the autophagic/lysosomal pathway, compensates this cellular phenotype. Transfection of siRNAs has been performed as described previously, 5 using siRNA s23340 (Ambion) for VPS13A inhibition, s29543 (Ambion) for VPS13C inhibition and 2-4390846 (Ambion) as a siRNA control.

| CRISPR/Cas9 techniques
Three guide RNA expression plasmids were tested in T7 endonuclease 1 mismatch detection assays 15 and the one that gave the best result was selected. For clonal isolation, the Cas9 and guide RNA expression plasmids were transfected in HeLa cells with Lipofectamine 2000 according to manufacturer's instructions. After 48 h posttransfection, cells were disaggregated and serial dilutions were performed up to the limit dilution of 5 cells/mL, and 100 μL (0.5 cells/ well) was cultured in 96-well plates (Falcon, 353072). Optical microscopy was used to monitor the appearance of single growth foci (clones). The identified clones were disaggregated and plated in a new 96-well plate (pass 1). When the cells reached 80%-90% confluence, they were transferred to a 24-well plate (pass 2) and then to a 6-well plate (pass 3). When they were again confluent, they were disaggregated and resuspended in 1 mL of fresh medium (DMEM).
Part of these cells were left for further growth and study, while the remainder was washed with 1X PBS and used for genomic DNA ex-  The LAMP1 signal ratio (A1/A2) is used as a measure of the level of accumulation. For this, cells were virtually divided into two halves and the area covered by the LAMP1 signal was measured using ImageJ (the larger signal was always assigned to A1).

| Rapamycin treatment
Cells were incubated with rapamycin (Calbiochem, 553210) at the indicated concentrations for 48 h before being fixed for immunofluorescence. For CRISPR/Cas9 VPS13A KO pool analysis, cells were treated with 100 nM rapamycin or equal volume of DMSO (dimethyl sulfoxide) 4 days after transfection with the guide RNA and Cas9 plasmids as described above. After 48 h of treatment, immunofluorescence was performed as described earlier.

| Data and image analysis
Image analysis was performed with ImageJ software. Data ob-

| Clonal isolation of VPS13A edited HeLa cells using CRISPR/Cas9 results in the selection of clones with residual level of VPS13A expression
Our first objective was to isolate VPS13A KO clones in HeLa cells using the CRISPR/Cas9 technique. A guide sgRNA targeting a sequence located in exon 1 was designed ( Figure 1A). Several clones were isolated by limiting dilution after transfection with the CRISPR/Cas9 plasmids. Genomic DNA was prepared from 11 clones, and PCR was performed using oligonucleotides flanking the target ( Figure 1B). While a single PCR product of the expected size (348 bp) was detected in HeLa WT cells, all clones tested yielded additional bands of larger size, indicative of large insertions in some of the 3 alleles of the VPS13A gene in HeLa cells ( Figure 1B). In addition, a PCR product migrating close to WT size was also detected in most clones. To verify that the corresponding allele also contained a deletion/insertion, this band was isolated and DNA was sequenced for 10 of the 11 clones. Half of the clones (6,9,17,23,28) contained a short deletion of 3, 6 or 12 base pair that maintained the open reading frame. In contrast, the other half (16, 18, 21, 22, 27) contained a two base pair deletion that disrupted the reading frame, with a one base shift in clone 18 ( Figure 1C Figure 1D). In these clones, we initially observed a marked growth defect that progressively recovered with each passage. After 15 days of growth, we rechecked the level of VPS13A by WB and, surprisingly, found low but detectable levels of VPS13A ( Figure 1E), as demonstrated by its sensitivity to siRNA treatment ( Figure 1F). These observations indicate that the isolated clones still had detectable levels of VPS13A expression despite successful editing of all three alleles.
Interestingly, although the resulting deletions of clones 16, 18, 21, 22 disrupt the reading frame, their editing also generates a potential alternative initiation codon in phase with the VPS13A ORF that would result in a truncated protein lacking the first 19 amino acids ( Figure 1C). Our results suggest that loss of VPS13A in HeLa cells impairs growth, which consequently favours the selection of clones that, although fully edited, have a residual level of VPS13A expression. The presence of residual levels of functional proteins in fully edited clones following CRISPR/Cas9 has been previously described and has been found to be more frequent than expected. 16 To avoid this, we set out to optimize a CRISPR/Cas9-based method for VPS13A that would eliminate the clonal selection step.

| A CRISPR/Cas9-based cell pool technique enables detection of accumulated late endosomes/ lysosomes in VPS13A-deficient HeLa cells
We reasoned that pooled cells taken shortly after transfection with CRISPR/Cas9 plasmids would contain a mixture of WT and edited cells not subjected to the selective long-term growth pressure required for clonal isolation. As expected, clusters of VPS13A positive cells were intermixed with non-expressing cells 6 days after transfection with the CRISPR/cas9 plasmids (Figure 2A immunofluorescence in this pooled population allowed us to discriminate cells expressing VPS13A (VPS13A+) from edited cells with undetectable VPS13A expression (VPS13A−). The specificity of the antibody in immunofluorescence was assessed using VPS13A or VPS13C siRNA-depleted cells ( Figure S1B). Under these conditions, the antibody was highly specific for VPS13A, as the signal obtained in VPS13A-depleted cells was almost undetectable, and the fluorescence pattern and intensity were not affected by VPS13C depletion. A limitation of this technique is that cells with low levels of VPS13A that might result from one unedited allele or from indels that maintain the reading frame might not give sufficient signal on immunofluorescence, and thus be classified as VPS13A−. Therefore, in addition to cells with complete absence of VPS13A, the VPS13A− pool may contain cells with low levels of VPS13A below the level of detection by immunofluorescence.
We performed simultaneous detection of endosomal markers and VPS13A to compare the distribution of these markers in VPS13A+ and VPS13A− subpopulations in the same microscopic preparation ( Figure 2B). Most VPS13A− cells showed some level of accumulation of late endosomes (Rab7A) and lysosomes (LAMP1), whereas the early endosome marker EEA1 showed a scattered distribution in both VPS13A+ and VPS13A− cells ( Figure 2B). These results are in agreement with previous data using VPS13A siRNA depletion in HeLa cells. 5,10 Late endosomes and autophagosomes move toward the perinuclear region as they mature to fuse with lysosomes that normally reside in this region. 17  measure of the asymmetric distribution of this marker ( Figure 2C).
As expected, we observed an increased LAMP1 accumulation ratio in VPS13A− compared to VPS13A+ cells, which is consistent with the previously described phenotype. 5

| The autophagic/lysosomal pathway may be a potential therapeutic target in ChAc
We hypothesized that the use of mTORC1 inhibitor rapamycin, a potent activator of lysosomal biogenesis and autophagy, [18][19][20] could rescue this phenotype. Figure 3A and autophagy partially compensates for the lysosomal dysfunction due to lack of VPS13A. However, it should be noted that rapamycin, through its inhibition of mTOR, may regulate pathways unrelated to lysosomal function, such as those related to insulin and amino acid signalling, which affect different aspects of cellular metabolism. 21 We cannot rule out that other lysosomeindependent pathways may play an indirect role in the observed lysosomal phenotype. This potential effect of rapamycin suggests a possible therapeutic benefit that should be studied in more detail in other models closer to the disease, such as the mouse models developed by other research groups. 14,22,23 Interestingly, rapamycin has been proposed as a possible treatment in other more prevalent neurodegenerative diseases such as Alzheimer's, Parkinson's and Huntington's diseases.
In summary, we have established a new simple cellular model of VPS13A deficiency based on CRISPR/cas9-modified HeLa cells.
Detailed characterization of CRISPR/Cas9 clones of VPS13A suggests that VPS13A inactivation impairs cell growth in this cell type, which generates a selective pressure to maintain residual levels of VPS13A protein in isolated clones. Very low levels of VPS13A appear to be sufficient to prevent the observed defects in lysosomal accumulation in this cell type. The unexpected sensitivity of HeLa cells to the lack of VPS13A is an experimental advantage that justifies the use of this model system to study VPS13A function but calls for caution when analysing phenotypes in isolated clones. To overcome the problem of clonal isolation, we use pooled cells after transfection, a convenient way to study subtle phenotypes due to the presence of an internal control, and which avoids the problem associated with studying genes that affect cell viability. Furthermore, this is the first time that rapamycin has been used in a preclinical study in the context of ChAc, and our results provide the first evidence suggesting that modulation of the autophagic/lysosomal pathway by rapamycin (or other TOR-dependent pathways) may be beneficial.
This finding warrants further investigation in more complex models on the potential use of rapamycin and the pathways affected by this treatment.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflict of interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.